By Sarah Bahari
A deadly influenza outbreak is spreading across the state. A homeless shelter is reporting a worrisome spike in cases of tuberculosis. A dangerous disease such as anthrax is intentionally delivered to a jam-packed arena.
In such scenarios, public health responders would have little time to answer questions. Where and how rapidly does the disease spread? Should public transportation be limited? In a shortage, who should be vaccinated? Researchers at the newly formed Center for Computational Epidemiology and Response Analysis at the University of North Texas are trying to answer such questions before a public health emergency actually occurs.
Their goal is to develop tools that could help us better understand — and ultimately respond to — infectious disease outbreaks. Using high-performance computers, students and faculty are creating predictive models that simulate disease outbreaks and help public health officials respond.
“How will a disease play out? This is the central question,” says Armin Mikler, associate professor of computer science and engineering and director of the center. “In the health field, we use computers to facilitate experiments through simulation. This allows researchers to conduct a ‘what if” analysis to explore multiple plausible scenarios.”
The center formed just months before the importance — and urgency — of the type of research it would tackle became evident in spring 2009, when swine flu became an international threat. Little information was available on the illness, and patients flooded doctors’ offices while entire school districts closed. As health officials across the country scrambled to contain the spread of the H1N1 virus, UNT researchers were acquiring state-of-the-art equipment and building the center’s infrastructure to help health professionals better understand such situations.
Jorge Reyes and Courtney Corley, both doctoral students in computer science and engineering who were working in UNT’s new research center, turned to social media to track swine flu and try to map its spread around Mexico and into the United States. Working with a company that tracks social media, Reyes and Corley combed through English- and Spanish-language blogs since October 2008 for all mentions of “flu” or “swine flu.”
The idea is that if people have the flu or know someone with the flu, they will mention it on a blog, Facebook or other social media outlet, Corley says. The students previously had tracked seasonal flu outbreaks using social media and found that their work closely mirrored the tracking done by the Centers for Disease Control and Prevention.
“Social media gives us one more way to track the flu virus,” says Corley, who graduated from UNT and is now a post-doctoral researcher at Pacific Northwest National Laboratory in Richland, Wash. “It’s far less expensive and time consuming than traditional methods of reporting. And if we catch something early, steps could be taken to prevent a catastrophic epidemic.”
Health experts say the center will provide invaluable tools in cases such as this.
“This sort of work could have reverberations across the health field,” says Karan Singh, professor and chair of the Department of Biostatistics at the UNT Health Science Center at Fort Worth and a member of the center. “This research has the potential to recognize the disease spread pattern and ensure strategies to provide timely interventions to contain and control an outbreak.”
The nerve center for the research is on the top floor of UNT’s Environmental Education, Science and Technology Building. This is what the researchers call the simulation chamber.
The centerpiece of the chamber is an enormous television screen — about 20 by 6.5 feet — used to display maps of cities or counties, diagrams of buildings and close-ups of neighborhoods. Cases of a disease and their geographic location may be represented by dots. As the disease spreads, the dots multiply and fan out.
Based on volumes of biological, demographic and environmental data such as housing figures, income, race, traffic patterns and average seasonal temperatures, the scientists eventually will be able to predict in populations large and small what groups are at risk, how the disease will spread and about how many people could be affected.
For instance, when a Fort Worth homeless shelter saw a spike in cases of tuberculosis several years ago, Joseph Oppong, geography professor and member of the center, and a group of students visited the shelter to get a feel for how the residents live.
Oppong and the students noticed that some of the residents were transient. They came and went, staying a night or two each week at the shelter, usually sleeping on a mat on the floor. But others clearly lived there — their lockers were decorated and they slept in beds with sheets.
At UNT, the researchers developed a computer simulation that detailed the interior of the homeless shelter, the placement of beds, restrooms and eating areas. A pattern emerged. Because the transients had fewer cases of tuberculosis than the live-in residents, a Health Science Center professor recommended replacing light bulbs with ultraviolet lights, which kill bacteria.
Recently, UNT researchers broadened their research to an entire Texas county. County officials asked for help assessing the response plan in case of a biological emergency, such as an anthrax or smallpox attack. In this sort of attack, the first 48 hours are crucial to containment.
Using census data, maps of highways and roads, traffic patterns, locations of schools and government buildings and a slew of other information, researchers developed models to analyze potential clinic locations, or points of distribution for vaccines. Supported by computational models, researchers selected several points of distribution across the county they believe would serve the highest number of people in the shortest amount of time to help control the spread of the disease.
“If an attack of this sort occurs, responders will have a very short time frame to respond. If plans are not formed and analyzed until after the emergency begins, it will be too late,” says Marty O’Neill II, a UNT doctoral student in computer science who worked on the project. “Responders need to know that the transportation network is capable of supporting timely travel to the points of distribution and the number of people who must be treated per minute.”
To execute these complex calculations, the simulation chamber at UNT needs an extraordinary amount of computational power. A new computing cluster that will power the system has been installed at Discovery Park, UNT’s nearly 290-acre research park north of the main campus. Portable visualization systems to view the simulation chamber operations remotely are housed at Discovery Park and at the Health Science Center, which has partnered with UNT on the center.
While the simulation chamber and center are new, they are both the culmination of nearly five years of research at UNT. The collaborative effort began when Mikler, in computer science and engineering, first wondered if there was a way to use computational techniques to help address public health problems. He brought in Oppong, who specializes in medical geography.
Medical geography, Oppong says, is the realization that where you live affects your health and the diseases to which you are most vulnerable.
Oppong would provide geographic data, and Mikler would create the software to simulate an outbreak. Before long, the professors realized they were missing a crucial angle: dynamic environmental factors.
Sam Atkinson, biology professor and director of UNT’s Institute of Applied Sciences, joined the partnership to provide the environmental expertise.
Developing a feasible vaccination strategy works only if scientists have the most up-to-date information on population, new roads, shopping centers and subdivisions, Atkinson says. For example, some diseases are highly dependent on land cover. Malaria is spread by mosquitoes, while avian flu migrates with bird behavior.
Once the disease enters the human population, other environmental factors come into play, Atkinson says. In the case of the flu, the environmental factors are endless: How many people do you work with or live with? What is the quality of air where you live? How often do you go to the grocery store?
“Human interaction is a huge factor in determining how a disease spreads,” Atkinson says. “We have to take all these factors into account in order to model disease outbreaks adequately enough to allow public health planners to begin understanding how various responses alter the patterns of an epidemic.”
The three researchers formed the Computational Epidemiology Research Laboratory and began recruiting additional faculty at UNT and at the Health Science Center. The laboratory eventually turned into the center, and the team has begun attracting external funding to support its research, including funds from the U.S. Department of Health and Human Services to purchase equipment.
Researchers will focus on infectious diseases for now, but they could expand the center’s mission to include chronic diseases, such as diabetes and cancer.
“We see endless possibilities with computational modeling,” Mikler says. “This research could dramatically enhance and improve our knowledge of how, why and where diseases spread.”
Studying Aircraft Fatigue
Experts across disciplines at UNT will help improve materials on aging aircraft for the military.
By James Naples
UNT leads NSF center creating cutting-edge software for complex networks.
By Sarah Bahari
Students in the early music program collaborate on musical masterpieces.
By Ellen Rossetti
Undergraduates experience the work of professional scholars with fellowship initiative.
By Nancy Kolsti
Librarians harvest defunct government web sites for UNT’s CyberCemetery.
By Alyssa Yancey
Instruments measure molecular fingerprints 1,000 times faster than before.
By Mellina Stucky
"Leading the pack" for innovative changes
National research status, new senior faculty
Top engineers, chemists, Fulbright specialists
New clusters, plant defense, Nature research
Experts in children’s issues, bioinformatics, interactive performing, mesoscale technology
Exploring tribal archives, corporate fraud, nanoscale circuits, international politics
Spanish language media, math, Arab art
Investing in research
Web page last updated or revised: January 27, 2010
Questions or comments about this web site? firstname.lastname@example.org
"University of North Texas," "UNT," "Discover the power of ideas" and their associated identity marks, as well as the eagle and talon graphic marks, are official trademarks of the University of North Texas; their use by others is legally restricted. If you have questions about using any of these marks, please contact the Division of University Relations, Communications and Marketing at (940) 565-2108 or e-mail email@example.com.